anthropic-academy-course-completion
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes browser automation (CDP/evaluate) to navigate the Anthropic Academy website, interact with course elements, and execute JavaScript callbacks such as
lessonPlayerCompleteCallback()to trigger lesson completion. - [PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection by design, as it ingests and processes external content from web pages to determine agent actions.
- Ingestion points: Text retrieved from quiz questions and options via DOM selectors (e.g.,
.question-text p,#sj-quiz-question-text p) onanthropic.skilljar.com. - Boundary markers: Absent. There are no instructions or delimiters defined to prevent the agent from following malicious instructions that could theoretically be embedded in course content.
- Capability inventory: The skill possesses browser automation capabilities, including navigation, element selection, clicking, and arbitrary JavaScript execution in the browser context.
- Sanitization: Absent. The skill does not specify any validation or filtering of the content retrieved from the external site before it is used by the agent to reason about answers.
Audit Metadata